摘要
瞬变电磁响应受到发射波形的影响,正演计算中需考虑发射波形.基于位移逆Krylov子空间投影的模型降阶算法能够精确模拟三维全波形瞬变电磁正演响应,但该算法计算精确的on-time响应非常耗时.目前实际工作中多数情况下都是对off-time时间段的数据进行处理解释,因此可以将正演问题简化为精确模拟考虑发射波形的off-time时间段的响应.本文实现了一种改进的基于位移逆Krylov子空间投影的模型降阶算法,能够实现任意发射波形的瞬变电磁off-time时间段的响应的三维快速正演.基于频谱分析理论实现了一种定量化的评价on-time采样波形精度的方法,能够在保证off-time响应精度的同时,极大的减少on-time波形采样点的数量,从而显著的减少正演计算时间.典型的半正弦波、梯形波和复杂的VTEM波的模型的正演结果表明,相比密集采样的正演结果,本文算法在保证off-time响应精度的同时,能够实现5倍以上的正演加速.
Shift-and-Invert(SAI)Krylov subspace technique has proved accuracy for the modeling of full-time time-domain electromagnetic(TEM)data.However,this method is expensive for computing on-time response.The solution time is roughly proportional to the number of on-time sampling time points.Fortunately,for the most part,only accurate off-time data is used for practice.In this paper,we present a modified SAI Krylov subspace method for quickly solving off-time TEM response.Based on spectrum analysis,we propose a quantitative method to evaluate the accuracy of on-time sampled waveform,and greatly reduce the number of on-time sampling time points.Numerical results for typical transmitting waveforms,such as half-sine,trapezoid and versatile-time-domain-electromagnetic(VTEM),demonstrate that the novel algorithm is accuracy for off-time TEM modeling,and more than 5 times faster than previously proposed schemes.
作者
周建美
刘文韬
齐彦福
李貅
戚志鹏
鲁凯亮
刘正银
邢涛
ZHOU JianMei;LIU WenTao;QI YanFu;LI Xiu;QI ZhiPeng;LU KaiLiang;LIU ZhengYin;XING Tao(College of Geology Engineering and Geomatics,Chang′an University,Xi′an 710054,China;Integrated Geophysical Simulation Lab of Chang′an University(Key Laboratory of Chinese Geophysical Society),Xi′an 710054,China;Shandong Provincial Transportation Planning and Design Institute,Jinan 250031,China;Beijing Tanchuang Resources Technology Co.,Ltd,Beijing 100071,China)
出处
《地球物理学报》
SCIE
EI
CAS
CSCD
北大核心
2022年第3期1160-1174,共15页
Chinese Journal of Geophysics
基金
国家自然科学基金项目(41704108,41830101)
长安大学中央高校基本科研业务费专项资金(300102261201)联合资助。
关键词
瞬变电磁
三维正演
发射波形
模型降阶
Transient electromagnetic
3D forward modeling
Transmitting waveform
Model order reduction